Match Selection and Refinement for Highly Accurate Two-View Structure from Motion

Zhe Liu 1, 2, 3 Pascal Monasse 1, 2, 3 Renaud Marlet 1, 2, 3
3 IMAGINE [Marne-la-Vallée]
CSTB - Centre Scientifique et Technique du Bâtiment, LIGM - Laboratoire d'Informatique Gaspard-Monge, ENPC - École des Ponts ParisTech
Abstract : We present an approach to enhance the accuracy of structure from motion (SfM) in the two-view case. We first answer the question: " fewer data with higher accuracy, or more data with less accuracy? " For this, we establish a relation between SfM errors and a function of the number of matches and their epipolar errors. Using an accuracy estima-tor of individual matches, we then propose a method to select a subset of matches that has a good quality vs. quantity compromise. We also propose a variant of least squares matching to refine match locations based on a focused grid and a multi-scale exploration. Experiments show that both selection and refinement contribute independently to a better accuracy. Their combination reduces errors by a factor of 1.1 to 2.0 for rotation, and 1.6 to 3.8 for translation.
Type de document :
Communication dans un congrès
ECCV 2014 ( European Conference on Computer Vision 2014), Sep 2014, Zurich, Switzerland. pp.818-833, 2014, Computer Vision – ECCV 2014 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings. 〈10.1007/978-3-319-10605-2_53〉
Liste complète des métadonnées

https://hal-enpc.archives-ouvertes.fr/hal-01153274
Contributeur : Pascal Monasse <>
Soumis le : mercredi 4 juillet 2018 - 12:10:36
Dernière modification le : lundi 16 juillet 2018 - 11:11:21

Identifiants

Citation

Zhe Liu, Pascal Monasse, Renaud Marlet. Match Selection and Refinement for Highly Accurate Two-View Structure from Motion. ECCV 2014 ( European Conference on Computer Vision 2014), Sep 2014, Zurich, Switzerland. pp.818-833, 2014, Computer Vision – ECCV 2014 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings. 〈10.1007/978-3-319-10605-2_53〉. 〈hal-01153274〉

Partager

Métriques

Consultations de la notice

264

Téléchargements de fichiers

185